Quantifying zooplankton species : use of richness estimators

Richness estimators (Jackknife 1, Bootstrap, Chao 1 and ACE) were used to relate zooplankton species richness with amount of water collected per sample and number of samples throughout the year for the limnetic region of Sapucai River compartment of Furnas reservoir, state of Minas Gerais, Brazil. Seven 100 L samples were collected in sequence using a motor pump, and seven 70 L samples were collected in sequence using a plankton net (68 μm mesh size) in vertical hauls, to totalize 450 L, in three stations of the reservoir. Twelve monthly samplings were carried out over a year. The assessment of richness was made by analyzing the asymptotic behavior of the estimator curves. The samplings reached the asymptote from 350 L of collection with trawls and 400 L using a suction motor pump and reached the plateau on the 8th collection, which included both dry and rainy seasons. Regardless of the type of sampling, the volume of 400 L and eight sessions throughout the year is enough to register 90% of the zooplankton richness in the environment.

Biological diversity can be understood at three levels: species, habitat and genetic diversity.Diversity indices use the density and richness of species and infer on the distribution equitability of species in the environment (Begon et al., 2007;Melo, 2008).
Knowledge of aquatic diversity is fundamental to implement conservation programs and rational use of resources, to assess environmental impact for licensing of hydroelectric power plants and to regulate fi sh farm areas (King & Porter, 2005;Eskinazi-Sant'Anna et al., 2005;Santos, 2006;Pinto-Coelho, 2004;Rocha et al., 2010;Santos-Wisniewski et al., 2011;Brito et al., 2011).These inventories become important because the degradation of natural ecosystems and biodiversity loss have increased in recent years.
To identify and count every species of a study area is very complex.Ecological studies search to estimate the species richness of these environments by sub-samples and the use of statistical techniques (Muirhead et al., 2006;Melo, 2008).Hence, the standardization of procedures and sampling eff ort is indispensable to reduce uncertainties and to make possible the comparison of species richness between diff erent study environments (Gotelli & Cowell, 2001).Among statistical techniques, non-parametric richness estimators, such as Chao, ACE and Jackknife estimators, have been used to improve sampling.These are based on the occurrence of rare species that appear in few samples or in low density (Magurran, 2011).Other estimators, such as Bootstrap, give the same value for every species collected, whether they are rare or common, to estimate the total richness (Santos, 2006).
The methodology and equipment used in samplings can interfere in species richness measurements, for example, the mesh size of the net used to collect aquatic organisms should consider organism size and escape capacity.In the same way, the collection procedure, such as the use of suction pumps or vertical hauls by net, integrating or not the water column, can interfere in results (Pinto-Coelho, 2004).Furthermore, the water volume for sampling should be enough to represent the environment diversity.For the aquatic environment, the standardization of sampling effort means that a minimal amount of collected water is established, per sample, for about 90% of the total environment richness to be registered (Heck et al., 1975).
Species accumulation curves are good tools for assessing the effectiveness of a sampling method, since they represent the cumulative number of species observed in an area (or volume) as a function of sampling effort (King & Porter, 2005;Muirhead et al., 2006).On the other hand, rarefaction curves are used for direct comparisons between populations, or samples, to obtain the number of species expected in a random sample (Magurran, 2011).
Due to seasonal variations in zooplankton composition and richness (Dumont & Segers, 1996), a preliminary sampling is recommended to determine which sampling effort is required to access the highest possible richness in tropical aquatic environments, especially in temporary ponds.In the tropics, the high predation rates by planktivorous fish, cyanobacterial blooms and high pollution due to anthropogenic activities, are factors that influence the loss of diversity and have significant effects in water bodies (Sarma et al., 2005).
This study aims to determine the minimum water volume needed per sample collected for the largest possible number of zooplankton species in the limnetic region of Sapucaí River compartment of Furnas reservoir, and the sufficient number of monthly samples over a year to reach maximum richness.

MATERIALS AND METHODS
The reservoir of the Hydroelectric Power Plant (HPP) of Furnas is located in the south of the state of Minas Gerais, Brazil.It has a flooded area of 1,450 km² and 250 km of length, in both of their two main sub-axes or compartments, the Sapucaí and Grande rivers, respectively (Corgosinho & Pinto-Coelho, 2006).The average depth is 13 m and the maximum reaches 90 m, near the dam.
Samplings were carried out in HPP Furnas reservoir at the junction of rivers Verde and Sapucai (VSJ) (21°27'03"S and 45°40'24"W) in December 2009, with a plankton net of 68 μm mesh size.First, seven samples of 100 L were collected sequentially integrating one-meter of water column, using a suction motor pump.After this, seven other samples of 70 L were collected with vertical hauls, using constant speed.Both methods were limited to onemeter depth from the water surface.The organisms were subjected to narcotization process with CO 2 saturation by the addition of carbonated water to avoid body contraction of the zooplankton individuals.The samples were maintained in polyethylene bottles and fixed with formalin 4% (v/v).Additional single samples of 400 L were collected monthly, in Barranco Alto region in this same reservoir, from March 2011 to February 2012.They were collected at stations BA1 (21°10'33"S; 46°00'51"W), BA2 (21°10'17"S; 46°00'38"W) and BA3 (21°10'04"S; 46°00'26"W).The suction motor pump was used to integrate the water column, from the station depth to the water surface, and the sample was prepared and maintained as above.
From the sample analysis the non-parametric estimators (ACE, Chao1, Jackknife1 and Bootstrap) of species richness were calculated based on abundance using the EstimateS 8.2 software (Colwell, 2009).Generally, these estimators make comparisons between rare species (uniques or singletons), present in a unique sample, and species present in, at least, two samples (doubletons or duplicates).These estimates were chosen according to King & Porter (2005) and Sousa et al. (2014).
The richness evaluation was done by asymptotic behavior analysis of curves obtained from richness estimators.

RESULTS
Firstly, the analysis of samples from the VSJ station (unique sampling data) was carried out.The curves of ACE, Chao 1, Jackknife 1 and Bootstrap estimators, as well as the observed richness curve (Sobs), for the two types of samplings (vertical hauls with mesh and suction motor pump) both adjusted for increasing volume using the seven collected samples, in sequence, showed that 90% of the maximum richness was reached, on average, with up to 400 L of filtered water (see Figs 2, 3).
For the suction pump sampling, the asymptote was obtained by filtration of 600 L. Only the Jackknife 1 estimator curve showed the organism richness increasing up to 700 L of sample, given that the curve of uniques began to decrease again (Fig. 2).For vertical haul sampling, about 80% of the maximum richness was reached using up to 140 L and above 95% using up to 350 L, for every richness curve.The observation of estimator curves showed that the asymptote was reached using up to 420 L (Fig. 3).
The rare species Tricocherca bicristata, Beucampiella sp., Chydorus eurynotus and Ascomorpha sp., as well as Lecane cornuta, Ptygura libera, Brachionus calyciflorus, Iliocryptus spinifer and Alonella lineolata were recorded in only one of the seven samples, being the earliest obtained from vertical hauls and the latest from suction pump sampling.
For analysis in time scale, twelve samples were collected every month for one year.Standard species accumulation curves were similar for the three sampling stations BA1, BA2 and BA3, showed by Figures 3, 5 and 6.Chao 1 and ACE estimator curves approximately overlapped with the Sobs curve, because there were no singletons or doubletons in the samples and species occurred in more than 10 ind.L -1 .Species with low frequency (uniques and duplicates) were recorded at high density, so Jackknife 1 and Bootstrap estimator curves showed species richness above those of the Sobs curve, in all samples.The minimal sampling effort was reached, for all curves, on the 8 th sampling (asymptote), for both dry and wet seasons samplings.

DISCUSSION
The two tested sampling methods, suction pump and vertical hauls, showed similar results and included in the inventory most of the species already registered for this reservoir, thus the two sampling methods are comparable.At 400 L water volume about 90% of the environmental species had been accessed for samplings using suction pump and a similar percentage was recorded with 350 L for samplings with vertical hauls using plankton net.It is considered a satisfactory inventory of species when between 50 and 75% of the species which could potentially occur in the environment are registered and the more frequently found species (or common) should be included in this percentage (Heck et al., 1975).
The greater species richness was recorded for samplings using vertical hauls with a plankton net.Rotifers were the most specious in this type of sampling while cladocerans were more representative with suction motor pump samplings.The most sensitive organisms, such as rotifers, may be damaged (including breakup and destruction of some individuals) during suction pump samplings, which makes their identifi cation diffi cult (Kozlowsky-Suzuki & Bozelli, 1998;Pinto-Coelho, 2004).On the other hand, the suction motor pump samplings select organisms with reduced escape capability and low (and slow) swimming movements, such as Cladocera (Pinto-Coelho, 2004).
The species richness for the three stations was very similar regarding the sampling throughout the year.With increasing sampling eff ort, the rarest species were identifi ed.However, when the sampling eff ort is excessive, errant species are recorded, contributing to an increase in the uniques curve and even in this case the estimators extrapolate the total richness of the environment (Magurran, 2011).Thus, it is believed that in 20 rare species sampled; only 8% are truly rare (King & Porter, 2005).
Many of the species identifi ed in other studies in the Furnas reservoir were not recorded in this study, as they covered a larger area of this reservoir surrounding the limnetic and littoral regions and the two compartments (Rivers Grande and Sapucaí), resulting in greater richness.In the review by Santos-Wisniewski et al. (2011) for Cladocera fauna of Minas Gerais, 94 species were recorded for the state distributed in 88 water bodies.Of these, 62 species were identifi ed in the Furnas reservoir.Among the species not recorded in this study in the limnetic region were Bosmina longirostris, Bosmina tubicen, Daphnia ambigua, Simocephalus latirostris, Simocephalus serrulatus and Moina micrura.
Generally, the records of Bosmina longirostris for Brazil are dubious, because according to De Melo & Hebert (1994), the species occurs in North America and B. freyi in South America.In the present study and in earlier studies in the Furnas reservoir, only B. freyi was recorded.
Although the study was carried out in the limnetic region, typical species of the littoral region were recorded, such as Alona intermedia, Alona yara, Alona guttata, Camptocercus australis, Chydorus pubescens, Chydorus eurynotus, Ephemeroporus sp., Ilyocryptus spinifer and     Tab.I. List of occurrence of zooplankton species and Dajoz Constancy Index (DCI) for the seven samples (S) collected by suction pump at the VSJ station in Furnas reservoir, state of Minas Gerais, Brazil (Ct, constants; F, frequent; C, common; R, rare).
Tab. II.List of occurrence of zooplankton species and Dajoz Constancy Index (DCI) for the seven samples (S) collected with vertical hauls in VSJ station in Furnas reservoir, state of Minas Gerais, Brazil.(Ct, constants; F, frequent; C, common; R, rare).Macrothrix cf.elegans.According to Fernando (2002), the distinction between limnetic and littoral zooplankton is often not observed in the tropics, which explains the occurrence of these phytophylous organisms in the samples.
In the review by Eskinazi-Sant' Anna et al. (2005), 300 species of Rotifera were recorded in Minas Gerais, which only six of these species are registered for the Furnas reservoir and recorded in this study, what indicates that an update for rotifers in Minas Gerais is necessary.Segers & Dumont (1995) identified 102 species of rotifers in 20 points distributed in the Broa reservoir, including sampling in the limnetic and littoral regions of the reservoir, while in this study 42 species were recorded, which only covered the limnetic region.
Jackknife 1 and Bootstrap estimators curves showed richness above that verified by the observed richness curve (Sobs) throughout the study.This result is common when non-parametric estimators are used to estimate the richness of zooplankton species (Dumont & Segers, 1996;Muirhead et al., 2006;Sousa, 2014).The estimator richness obtained Tab.III.Cont.
by Jackknife 1 curve in BA1 station evidenced an asymptote, while for Bootstrap estimator curve, the asymptote was observed for BA1 and BA2 stations.The adjustments to other estimators curves showed a tendency to the formation of a asymptote, while for the uniques curves the asymptote was observed at all stations since 9 th sampling.
The estimator curves showed that only between 3% and 16% of environmental potential richness was not recorded and the sampling may be considered satisfactory.The richness species determined by the unique curves showed the asymptote beginning at 9 th successive monthly sampling.It indicates that the monthly sampling can be reduced to eight so that most of environment richness will be achieved in, in time scale.In the tropics, there is less variation in the seasonal succession of zooplankton species and species co-occur in both the summer and winter seasons (Dumont, 1994).Thus, a smaller amount of samples is necessary and the time and sampling costs could be reduced.
The results obtained with the volume collected showed that 400 L are sufficient to record 90% of the environment richness, regardless of the sampling type (vertical hauls with a plankton net or suction motor pump).Also, another recommendation is to decrease from twelve to eight the monthly collections throughout the year, to goal a zooplankton limnetic species inventory, as the maximum number of species of Cladocera and Rotifera will be obtained in the waters of the Furnas reservoir in eight months.

Fig 1 .
Fig 1. Sampling stations in the Hydroelectric Power Plant of Furnas reservoir, state of Minas Gerais, Brazil (A, Barranco Alto region; B, junction of rivers Verde and Sapucai -VSJ).

Fig. 2 .
Fig. 2. Species accumulation curves, uniques and duplicates for the VSJ station in Furnas reservoir, state of Minas Gerais, Brazil, collected by suction pump.

Fig. 3 .
Fig. 3. Species accumulation curves, uniques and duplicates for the VSJ station in Furnas reservoir, state of Minas Gerais, Brazil, collected with vertical hauls.

Fig. 4 .
Fig. 4. Species accumulation curves, uniques and duplicates for the BA1 station of Furnas reservoir, state of Minas Gerais, Brazil, from March 2011 to February 2012.

Fig. 5 .
Fig. 5. Species accumulation curves, uniques and duplicates for the BA2 station of Furnas reservoir, state of Minas Gerais, Brazil from March 2011 to February 2012.

Fig. 6 .
Fig. 6.Species accumulation curves, uniques and duplicates for the BA3 station of Furnas reservoir, state of Minas Gerais from March 2011 to February 2012.